Collaborative Research: AF: Small: New Directions in Algorithmic Replicability

Replicability and reproducibility in science are critical concerns. The fundamental requirement that scientific results and experiments be reproducible and replicable is central to the development and evolution of science. In recent years, these concerns have been even more amplified as several scientific disciplines are turning to data-driven research that enables exponential progress from data democratization and affordable computing resources. The replicability/reproducibility issue has received attention from a wide spectrum of entities ranging from general media publications to scientific publication venues to professional and scientific bodies. This project focuses on addressing the pivotal need for replicability and reproducibility in scientific research that involves randomized computations. The project will explore new directions in algorithmic replicability. The approach is threefold: (1) defining and formalizing new concepts of replicability across various computational contexts, (2) developing methodologies for the design of replicable algorithms, and (3) investigating the limitations and costs associated with replicability. This initiative seeks to contribute fundamentally to the field of replicable and reproducible computations. It focuses on the development of innovative algorithms and establishing lower bounds in replicability. The research is expected to have a significant impact on areas such as machine learning and large data analysis. Beyond these immediate fields, the principles and techniques devised under this project are anticipated to influence a broader spectrum of scientific and technological domains. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.